AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

GenieBlue: Integrating both Linguistic and Multimodal Capabilities for Large Language Models on Mobile Devices

Xudong Lu, Yinghao Chen, Renshou Wu, Haohao Gao, Xi Chen, Xue Yang, Xiangyu Zhao, Aojun Zhou, Fangyuan Li, Yafei Wen, Xiaoxin Chen, Shuai Ren and 1 more

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Unified any-modality language models · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add 2M pure-text samples to fine-tuning data (ablation)BlueLM-V-3B (BlueLM-3B) · MLLM benchmarksAVG accuracy: 66.19 -> 66.19No clear effect
add 2M pure-text samples to fine-tuning data (ablation)BlueLM-V-3B (BlueLM-3B) · LLM benchmarksAVG score: 43.78 -> 47.03Improved
add 2M pure-text samples to fine-tuning data (ablation)BlueLM-V-3B (BlueLM-3B) · AlignBenchscore: 3.87 -> 3.84No clear effect
add 2M pure-text samples to fine-tuning data (ablation)BlueLM-V-3B (Qwen2.5-3B) · LLM benchmarksAVG score: 49.29 -> 51.45Improved
add 2M pure-text samples to fine-tuning data (ablation)BlueLM-V-3B (Qwen2.5-3B) · MLLM benchmarksAVG accuracy: 65.01 -> 64.71No clear effect
replace LoRA with CogVLM-Skip visual experts (ablation)BlueLM-3B · MLLM benchmarksAVG accuracy: 60.78 -> 63.94Improved
replicate blocks at every quarter (Skip) instead of last quarter (Post) (ablation)GenieBlue (BlueLM-3B) · MLLM benchmarksAVG accuracy: 62.10 -> 65.52Improved
replicate blocks at every quarter (Skip) instead of first quarter (Pre) (ablation)GenieBlue (Qwen2.5-3B) · MLLM benchmarksAVG accuracy: 60.46 -> 63.26Improved
use GenieBlue-Skip instead of CogVLM-SkipBlueLM-3B · MLLM benchmarksAVG accuracy: 63.94 -> 65.52Improved
use GenieBlue-Skip instead of CogVLM-SkipQwen2.5-3B · MLLM benchmarksAVG accuracy: 62.87 -> 63.26Improved
use non-shared base deployment instead of shared base (ablation)GenieBlue-Skip (BlueLM-3B) · LLM benchmarksAVG score: 54.40 -> 60.16Improved
use non-shared base deployment instead of shared base (ablation)GenieBlue-Skip (Qwen2.5-3B) · LLM benchmarksAVG score: 52.57 -> 60.29Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy